Abstract Problem behaviors in demented patients, such as combativeness, wandering, and sundowning, are a major source of caregiver stress and a common reason for the institutionalization of the patient. Based on the well-accepted notion that these behavioral disturbances stem from agitation, we propose a novel approach of using electrodermal activity (EDA) to assess the degree of agitation in dementia patients. The goals are, first, to detect increasing agitation before it surfaces in the patient's behavior and, second, to devise methods of reducing the agitation to prevent the aberrant behavior from appearing. Almost no work has been done using EDA in dementia patients. The literature indicates that the latency and amplitude in dementia patients are similar to normal controls, implying that we can probe the individual's responses as with standard biofeedback techniques. Because of the paucity of data available, we will first develop a set of standard methods for studying EDA in dementia patients and monitoring spontaneous activity. Our group has developed a wearable device that will measure EDA and transmit it using Bluetooth. We will validate the device against the usual reference method, skin conductivity measured at the finger tips. Once validated, we can monitor long-term spontaneous EDA in dementia patients. This will allow the study of circadian rhythms in dementia patients, which is important in and of itself. These long term studies will aid in identifying EDA activity that may precede agitation. We will also study stimuli that decrease EDA, which the patient cannot do on his own as is required by standard biofeedback techniques and which may prevent the progression to problem agitation. If successful, this work could decrease caregiver burden and help dementia patients remain in their homes longer.